Atomistic Simulation of the Formation of Nanoporous Silica Films via Chemical Vapor Deposition

To date, membranes which could be used to separate gas mixtures at high temperatures (notably combustion or other process gas streams at temperatures T > 750K) have only been developed with limited success and it is apparent that this modest pace of growth is, in part, due to a lack of a detailed understanding of the membrane fabrication process at an atomistic level. One particular family of materials which has been shown to exhibit high permselectivities at elevated temperatures are thin films of amorphous silica glasses formed via chemical vapor deposition (CVD) techniques. In earlier work reported by this group [1-3] the possibility of generating membranes with both high kinetic selectivities for CO2 capture and acceptable permeability has been demonstrated. However, the influence of the experimental conditions employed during the CVD fabrication process on the fundamental microscopic details of the membranes formed still needs to be clarified. In this work, we investigate the creation of moderate density silica films via direct simulation of the CVD process at intermediate to high temperatures. To model the creation of nanoporous silica layers via CVD we apply a hybrid kinetic Monte-Carlo (KMC) method [4]. Lattice KMC [5] is used for the elementary reactions and an off-lattice method [6,7] is employed for silica network relaxation and bond switching moves. The sensitivity of the resulting layer structure to the substrate temperature, composition of the vapor, and the initial distribution of substrate seed sites is examined. The outcome of this work will assist in providing guidelines for the protocols needed to fabricate high temperature permselective membranes for the separation of CO2 from combustion gas mixtures. Kinetic Monte Carlo simulation of the CVD process The reactions which are deemed to take place are as follows. The deposition of Si(OH)4 occurs as |-O-H + Si(OH)4 ⎯ ⎯ → ⎯ F k1 |-O-Si(OH)3 + H2O (1F) and its reverse |-O-Si(OH)3 + H2O ⎯ ⎯ → ⎯ R k1 |-O-H + Si(OH)4 (1R) In addition the condensation (annihilation) reaction |-O-H + H-O-| ⎯ ⎯ → ⎯ F k2 |-O-| + H2O (2F) and its reverse |-O-| + H2O ⎯ ⎯ → ⎯ R k2 |-O-H + H-O-| (2R) take place as do the reversible switching reactions (see [6]), |-O-| + H-O-| ⎯ ⎯ ⎯ ⎯ ⎯ → ← = R k F k 3 3 |-O-H + |-O-| (3) |-O-| + |-O-| ⎯ ⎯ ⎯ ⎯ ⎯ → ← = R k F k 4 4 |-O-| + |-O-|. (4) These reactions allow silanol group and/or siloxane bridge mobility within the medium according to the Wooten, Winer, and Weaire formalism [8]. Hybrid KMC Algorithm The deposition and condensation reactions (1 and 2) are treated by a lattice KMC method and the switching reactions 3 and 4 (which are believed to occur much more frequently) are treated via standard Metropolis MC). The selection of which reaction event should be sampled at any given time is determined by the frequencies of the reactions. In general for each of the above reactions we can write and iF iF Ai Bi iR iR Ci Di r k c c r k c c = = with ⎟ ⎠ ⎞ ⎜ ⎝ ⎛− = kT E A k iF A iF iF , exp In the case of the forward reaction the frequency with which we observe the reaction of one molecule of component A is 1 iF iF F Vr t ν = = Δ Similar frequency expressions can be obtained for each of the forward and reverse reactions and in the lattice KMC algorithm one simply selects the reaction event to be considered with a probability